The Carolina Center for Exploratory Genetic Analysis will establish an interdisciplinary infrastructure for the efficient identification of the complex genetic traits underlying human diseases, based on clinical studies, population studies, and model systems. The planning stage for this center will build a collaborative community of investigators and deploy prototype infrastructure, driven by the quantitative analysis of relationships among genotypes and clinical or experimental phenotypes in three contexts: family linkage studies (susceptibility to alcoholic addiction), expression profile studies (breast cancer) and public health studies (atherosclerosis risk in communities). This effort will rely on the combined expertise of three complementary groups of UNC scientists: (a) experimental geneticists, (b) quantitative experts in statistics and biostatistics, and (c) computer scientists with expertise in algorithm development, software construction, and high-performance computing. To reduce the barriers between data providers and data analyzers, we will organize a series of intensive, specialized workshops, colloquia and intramural meetings. [unreadable] [unreadable] We believe the next major breakthroughs in our understanding of biology and disease will come from the integrated analysis of genetic data and its expression as phenotypes. To accommodate the diverse, multi-investigator data bases necessary to answer such questions, we will develop a prototype, extensible data model and provide access to data via a portal constructed using the Open Grid Computing Environment toolkit. To advance the practice of integrated analysis we will incorporate new methods under development into the data analysis workflow. These include new techniques in linkage analysis (oligogenic analysis, multivariate linkage analysis, epistasis, and genotype by environment interaction), subspace clustering, and association analysis (quantitative trait nucleotide analysis). For the interactive use of these computationally intensive techniques, we will explore new visualizations and develop high-performance implementations. [unreadable] [unreadable]